Research On New GVF Snake Model

The circuit board is one of the basic part of aircraft, its work has absolute influence on flight safety. Because the circuit board structure is complicated, the traditional contact type diagnosis can not meet the actual requirements of the probing mission. Therefore, the study of non intrusive diagnosis of airborne circuit board has important practical significance. The infrared thermal imaging technology is one of the key research which uses infrared thermal imaging system to get the image data to analyze and process the heating information of the circuit board. In the diagnosis process, the image segmentation is the first step which is also one of the key step. Because of the noise and the complex distribution of the circuit board infrared image, the selection of the segmentation algorithm has some special characteristics. After doing research of common image segmentation algorithms, the GVF Snake model is selected to study which has good image segmentation effect in other fields.At first, the traditional GVF Snake model and several improved new models in recent years are introduced, then the external force field and numerical implementation are analysed and simulation of circuit board infrared image segmentation are conducted. The simulation results show that the new model solve problems that the traditional model is sensitive to noise and has the incapability of moving to the deep boundary, but there still exists shortcoming that the convergence results are susceptible to interference when there is a strong boundary near the contour.The contour error convergence is defined as distraction problem, then a number of experiments are finished to analyse the causes of the problem and possible solutions are proposed to verify the effect. Edge map extraction of GVF Snake model is combined with improved canny operator successfully solve this problem. In addition, a new internal energy constraint is used to matches GVF field external energy, then a new subsection optimization GVF Snake model which combined with the greedy algorithm is proposed. The new model is different from traditional model and other improved models, it solves the problem of weak edges lost in the traditional GVF Snake model, improves the segmentation accuracy and practical value and reduces the amount of mathematical calculation. Different from the greedy algorithm, new model obtain the global optimal solution instead of local optimal solution.In addition, the improved model can also be applied to other image segmentation which has distraction problem, such as face and medical image segmentation. Image preprocessing of the new model is also a research direction in the future.